Outliers detection in the urban areas. Case studies: Romania, Bulgaria and Hungary

Abstract

Outliers are values that are questionable because they are much bigger or smaller than the majority of the values analyzed. The outlier detection is a research domain about which not a lot of persons are aware of, and because of that the identification of the abnormal values is often a skipped step in the analysis of a database. However, these atypical values which are generally ignored could result into interesting researches which could reveal valuable information about a particular area or about a particular phenomenon studied, exposing patterns which otherwise would have not been observed. Yet, a solid research should be done before removing the observations which were identified as being atypical and this paper emphasizes the duality of the term outlier and the necessity of performing an identification based on more than one method, before deciding if a suspicious value should or should not be eliminated from a research.